[sdiy] Query about best cap type for '2164 stability network

Mattias Rickardsson mr at analogue.org
Mon May 7 12:26:56 CEST 2018

On 5 May 2018 at 00:37, Neil Johnson <neil.johnson71 at gmail.com> wrote:

> > Ok, a follow-up question, since I’m here to learn.
> >
> > What’s the relationship between the 30K input resistor value and the
> > 500R/560pF lead-lag compensator values?
> >
> > The stated RC give a 1/2PiRC of about 570KHz. Where does 30K fit into the
> > picture? How did AD’s engineers arrive at those values? They provide
> them,
> > but they don’t tell you how they were derived or why, which would have
> been
> > more interesting. Perhaps to some of you it’s obvious (hence the
> omission in
> > the datasheet), but to me it’s totally obscure.
> Since ADI stopped making that chip some time ago, I doubt they will be
> able to answer now.  On the other hand I have asked SSI to provide
> details, but because it involves knowledge of the internal design of
> the 2164 I doubt we will get the exact equation.  I've had some
> discussions with SSI on coming up with compensation network values for
> different signal source impedances (we got down to around 10k IIRC)
> and I hope they put some of those figures into a revised datasheet.

SSI have tried out different combinations of values and I'm also hoping for
some more info and explanations about this in the datasheet. I've just
started using an RC stability network of 226R/1200p, which seems to be
their best suggestion so far for an input resistance of 15k with the
SSI2164 and the supply voltages used in this application.

If C0G is used preferred over X7R, the drawback of going up to 1200pF is
that they are getting larger and more expensive.

On 6 May 2018 at 20:09, <rsdio at audiobanshee.com> wrote:

> Don’t underestimate the effects of microphony. I once had a non-audio
> circuit with capacitors where tapping on the case would cause spikes in the
> control signals because I just picked the cheapest caps. These were in the
> 560 pF range, but I was actually trying various values within an order of
> magnitude.

Luckily I haven't encountered any microphony problems using X7R in a
million instances of these RC networks, but I agree that it's wisest to use

> So, you have to ask whether a non C0G/NP0 cap would alter the gain in the
> VCA and thus affect the audio, even if only slightly.

The RC network is on the audio input, not the gain input. So the spikes
would be added to the audio directly.

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://synth-diy.org/pipermail/synth-diy/attachments/20180507/51ac7f54/attachment.html>

More information about the Synth-diy mailing list